Predictors of Sustained Response to Functional Restoration in a Military Population

ABSTRACT

Introduction

The purpose of this study was to determine if improvement in pain impact and functional performance following a functional restoration (FR) program was sustained up to 6 months posttreatment and to identify predictors of sustained improvement.

Materials and Methods

Secondary analysis of data collected during randomized clinical trial. Study population included 108 US active duty service members who completed an FR program, as well as 3- and/or 6-month follow-up assessments. Primary outcome measure was the NIH Research Task Force (pain) impact score (PIS). Secondary outcome was a composite functional performance measure of treadmill, lifting, and carrying tolerances. Variables analyzed to determine their predictive value included demographics; treatment hours; measures of pain intensity, function, mood, sleep, social satisfaction, pain catastrophizing, kinesiophobia, self-efficacy, pain acceptance, patient activation, functional performance, and neuropathic pain.

Results

Mean PIS and functional performance improved significantly immediately following FR, but after 6 months, only improvement in functional performance sustained. Responder analysis showed that 6 months after FR, 42% of participants reported improvement that exceeded the minimal clinically important difference in PIS or functional performance. Predictors of sustained PIS improvement included younger age, absence of neuropathic pain features, less self-rated disability, better baseline functional performance, and worse baseline PIS. Predictors of sustained functional performance improvement included more treatment hours, lower baseline pain catastrophizing, and lower baseline functional performance.

Conclusions

This study supports the investment of treatment time in FR to yield sustained clinically meaningful improvement, as observed in over 40% of this study’s military participants. Pretreatment predictors of sustained response included lower pain catastrophizing and absence of neuropathic pain. Further research is needed to determine if treatments that improve pain catastrophizing and neuropathic pain will result in sustained improvement in pain impact and functional performance following FR program participation.

INTRODUCTION

Increasing access to interdisciplinary pain care is a priority of the US Department of Defense.¹ Defense Health Agency Procedural Instruction 6025.04 on pain management and opioid safety in the Military Health Care System endorses an interdisciplinary approach to pain management including standard rehabilitative care (SRC; physical therapy [PT] and occupational therapy [OT]), psychological care, and complementary and integrative health (CIH) therapies (chiropractic, acupuncture, yoga, and massage).² The Agency for Healthcare Research and Quality³ (AHRQ) defines multidisciplinary pain rehabilitative (MDR) programs as coordinated programs offering biopsychosocial treatment components provided by professionals from at least 2 different specialties. Interdisciplinary pain programs are MDR programs with a high degree of coordination between providers. Functional restoration (FR) programs are a type of interdisciplinary program that includes PT, OT, and psychosocial interaction using a sports medicine approach.⁴ At least 10 military pain management centers offer FR programs for the treatment of chronic musculoskeletal pain.

Although the effectiveness of FR programs is well established in civilian populations where they are predominantly related to workers’ compensation,³ there are limited reports of FR programs involving military populations. There are important differences between military and civilian workers’ compensation populations: military FR participants are overall younger, more physically conditioned, have fewer chronic medical conditions and are more likely to be men. To date, only one randomized clinical trial (RCT) compared FR to usual care in a US military population. In that study, the FR group had significantly better posttreatment outcomes across a spectrum of psychosocial and functional performance measures.⁴ Additionally, service members in the FR group showed significantly greater improvements in pain intensity, function, and emotional status after 6 months and reduced health care utilization at the 1-year follow-up.⁴

Over the past decade, 4 nonrandomized pretreatment versus posttreatment analyses of FR effectiveness in active duty service members were published. They all showed improvement in pain intensity, pain impact, and/or function following treatment,^5–8 and one identified predictors of posttreatment response.⁸ The effectiveness of FR has also been established in US veteran populations. Significant improvements across core outcome measures (including pain intensity, function, sleep, and pain catastrophizing) were reported in a nonrandomized study of treatment outcomes of more than 900 veterans who participated in FR programs at 5 different Veterans Health Administration (VHA) facilities.⁹

There is high clinical relevance in examining whether FR program participation leads to sustained improvements in pain outcomes and in identifying variables that predict sustained improvement. Knowledge of demographic, clinical, or treatment predictors could be used to recommend FR to individuals who are most likely to experience sustained clinically meaningful benefit. However, to date, only one study⁴ has assessed outcomes in military populations beyond the immediate posttreatment period, and there have been no published studies of predictors of sustained improvement following FR program participation in military populations. To address these gaps, this study aimed to determine if FR treatment response among active duty service members is sustained up to 6 months following treatment and to identify predictors of sustained improvement.

METHODS

The parent study protocol for this secondary analysis was approved by the US Army Regional Health Command-Pacific institutional review board after completion of an ethics review (protocol no. 215050), registered on ClinicalTrials.gov (NCT04656340), and was more fully described previously.⁸ The parent study was conducted between 2015 and 2020, and a total of 210 active duty Army, Navy, and Air Force service members consented to study participation. Treatment group assignment was not found to be an independent predictor of treatment response in the parent study. Thus, for the current study, both treatment groups were combined to evaluate predictors of sustained improvement.

SAMPLE AND INTERVENTION

To participate in the parent RCT, active duty service members were required to meet minimal functional thresholds including ability to stand up and sit down independently; walk or jog on a treadmill for 6 minutes; and complete at least 2 of the following: lift 20 pounds from floor to waist level, lift 20 pounds from waist to shoulder level, and/or carry 20 pounds a distance of 40 feet. A complete list of inclusion and exclusion criteria of provided in Appendix 1. Study participants were randomized to a 3-week program of either SRC alone (twice weekly PT, OT, and psychoeducation) or SRC in combination with CIH therapies (once or twice weekly chiropractic, acupuncture, and yoga). After 3 weeks of treatment, both treatment groups proceeded to an intensive FR program, which included 12 full days of treatment over 3-6 weeks. Each FR treatment day included 4 hours of physical activity, supervised by a PT or OT professional or a yoga therapist, and 1 hour each of cognitive behavioral therapy and educational classes. Study participants who randomized to SRC alone were given the option of engaging in the CIH program after completing FR (Fig. 1). Among the 42 study participants who randomized to SRC alone and also completed 3- and/or 6-month follow-up measures, nearly all (n = 39, 93%) elected to engage in CIH following FR. Therefore, this secondary analysis included pooled data collected from 2 subpopulations of study participants of approximately equal size: those who completed 3 weeks of CIH prior to FR and, with the exception of 3 study participants, those who completed 3-6 weeks of CIH following FR.

FIGURE 1.

Design of parent study.

Outcome Measures

Pain specialty clinics in the US Military Health System use the Pain Assessment Screening Tool and Outcomes Registry (PASTOR) to collect standardized patient-reported outcomes. To assess pain intensity, PASTOR includes the Defense and Veterans Pain Rating Scale (DVPRS), which uses a color-coded 11-point numeric rating scale of intensity (0-10), with descriptive anchors of pain severity for each rating.¹⁰ In addition, PASTOR includes a battery of Patient-Reported Outcome Measurement Information System (PROMIS) measures developed by the NIH to assess psychological, social, and functional well-being.¹¹

The primary outcome measure was the NIH Research Task Force (pain) impact score (PIS), which is a composite measure of average pain intensity, pain interference, and physical function.¹² Average pain intensity during the previous week was determined using DVPRS, and PROMIS measures of pain interference¹³ and physical function¹⁴ were determined by computing the raw score equivalent of the computer adaptive testing T-scores, each having a range of 4-20, for an overall PIS range of 8-50. The minimal clinically importance difference (MCID) of the PIS is estimated to be 3 based on the MCIDs of its component measures.¹⁵ The secondary outcome measure was a composite score of the treadmill-lift-carry (TLC) battery, a functional performance assessment developed by the study team. The TLC battery includes treadmill walking or jogging tolerance measured in minutes using a modified Naughton protocol, lifting tolerance (floor to waist and waist to shoulder lifts measured in pounds) and carrying tolerance (40-foot carry measured in pounds). A PT or OT clinician instructed study participants to continue to walk or jog on the treadmill and to gradually increase the magnitude of weight lifted or carried until they experienced an increase in pain intensity or reached maximally safe exertion. The metabolic equivalent score from the treadmill test and the amount of weight lifted or carried were converted to z-scores to place them on the same scale and then summed to create a composite score. The composite score was converted to a 0-100 T-scale score to increase interpretability (average score of 50 and a standard deviation of 10). The MCID of the TLC composite score was determined in a previous analysis to be an increase of 6 points.¹⁶

Additional Measures

Other PROMIS measures analyzed were depression, anxiety, anger, sleep-related impairment, fatigue, and satisfaction with social roles.¹¹ Established normative score ranges determined normal, average, and mild or low, moderate, and severe scores. PASTOR also includes the PROMIS neuropathic pain screen,¹¹ which includes 5 items that assess characteristics associated with neuropathic pain, such as pins and needles, tingling, stinging, electrical sensation, and numbness. With the exception of the neuropathic pain screen, all PROMIS measures in PASTOR use computerized adaptive testing, which reduces the participant response burden compared with original versions of the PROMIS assessment tools.

In addition to PASTOR, study participants completed selected supplemental questionnaires that have relevance to pain outcomes. The Pain Catastrophizing Scale (PCS)¹⁷ is a 13-item measure of a person’s tendency for (1) rumination, (2) magnification, and (3) helplessness. The PCS demonstrates excellent discriminant validity in the context of both experimental pain and clinical pain, with higher scores strongly correlated with negative pain-related thoughts, emotional distress, and greater perceived pain. The PCS was added to PASTOR in 2018, after data collection was completed for the current analysis. The Tampa Scale for Kinesiophobia (TSK-11)¹⁸ is an 11-item survey assessing pain-related fear in people with back pain. The TSK-11 has good internal consistency (Cronbach’s alpha = 0.79), test–retest reliability (intraclass correlation coefficient  = 0.81, standard error of measurement = 2.54), and responsiveness (standardized response means = -1.11). The Pain Self-Efficacy Questionnaire¹⁹ is a 10-item survey assessing people’s self-efficacy beliefs with respect to their pain and has excellent internal consistency (Cronbach’s alpha = 0.92), test–retest reliability (r = 0.73), and construct validity against other domains of health such as mental health, pain assessments, and coping behaviors. The Chronic Pain Acceptance Questionnaire (CPAQ-8)²⁰ is an 8-item survey of (1) the degree to which people engage in life activities regardless of pain and (2) their willingness to experience pain or the inverse of engaging in behaviors to limit pain. The CPAQ-8 has good internal consistency (Cronbach’s alpha = 0.77-0.89) and good construct validity against other domains of health such as depression, anxiety, pain interference, and pain severity. The 10-item Patient Activation Measure (PAM) measures patient activation²¹ by assessing the individual’s skill, knowledge, and confidence for managing health and health care. Each item has 4 response options: disagree strongly, disagree, agree, and agree strongly. Scores range from 0 to 100, with higher scores indicating higher patient activation. The PAM has moderate-strong internal reliability (a = 0.87).²¹ The Roland-Morris Disability Questionnaire (RMDQ-24)²² modified for use in people with general chronic pain conditions is a 24-item survey to assess self-rated physical disability caused by pain, including 20 items that measure activity limitations, 2 that measure psychological impairments, 1 that measures pain and symptoms, and 1 that measures sleep disturbances. The RMDQ-24 includes statements that represent routine activities that are completed or avoided and asks respondents to endorse those activities that describe their status today. A sum score of 0-24 counts the number of endorsed items, with higher scores indicating worse functioning. Measurement properties of the RMDQ-24 when completed by people with low back pain show sufficient test–retest reliability, construct validity, and responsiveness.²²

Demographic and Treatment Variables

Demographics for each study participant were obtained from PASTOR and Military Health System Management Analysis and Reporting Tool (M2) databases and included age, sex, rank group, income, educational level, and occupational category. Data on race and ethnicity were not available for analysis. Treatment variables accessed from the M2 database included the number, discipline, and duration of clinical encounters at the study site.

Statistical Analyses

The focus of the analysis was on the short-term and intermediate-term (3 and 6 months, respectively) outcomes post-FR. Therefore, analytical models were restricted to include only participants who had outcome data at 3 and/or 6 months. Demographic characteristics for participants who completed different types of measures (PASTOR questionnaires, supplemental questionnaires, and functional performance measures) at 3 and/or 6 months were compared to those of the full sample of the parent study using chi-square tests. Mixed regression models were used to assess change from baseline for all questionnaire and physical performance measures. Responder analyses were conducted by calculating the number and percentage of participants who met minimal clinically important differences for the primary and secondary outcome measures (change of 3 units for PIS and 6 units for composite functional performance score). Additionally, mixed regression models were used to determine predictors of sustained improvement in the primary and secondary outcome measures. Demographic variables, baseline questionnaire and functional performance measures, and treatment variables were assessed as predictors. A separate model was run for each predictor with an interaction term between the predictor and a categorical time variable. The outcome was the change in composite score (PIS or functional performance) from baseline for each time point. Time points included were post-FR, 3 and 6 months.

RESULTS

There were 210 participants enrolled in the parent study. Table I displays the number and demographics of participants who completed the PIS (n = 91) and the composite functional performance measure (n = 96) at 3- and/or 6-month follow-up. The distribution of variables in each follow-up population was statistically similar to that of the larger population enrolled in the parent study, with the exception of relationship status, with a higher proportion of those who completed follow-up functional performance measures reporting partnered status. The participants included in this analysis were predominantly male, active Army service members, younger than age 35, partnered, with at least some college education. Most did not have job-related activity restrictions issued by medical providers at the time of study enrollment, and most reported that their pain condition was not related to deployment. A comprehensive list of demographic and clinical factors analyzed is provided in Appendix 2.

TABLE I.

Demographics of Participants Enrolled in the Parent Study and Who Completed Pain Impact Score or Functional Performance Assessment 3 and/or 6 months Posttreatment

		Parent study		Completed 3- &/or 6 month measures
				Impact Score,		Functional Performance,
		n = 210		n = 91		n = 96
		n	%	n	%	n	%
Sex	Men	176	83.8	76	84.3	83	86.5
	Women	32	15.2	15	15.7	13	13.5
Age	≤35 yr	124	59.0	51	56.0	53	55.2
	>35 yr	84	40.0	40	44.0	43	44.8
Education	High school	49	23.3	21	23.6	22	23.2
	Some college +	154	73.3	68	76.4	73	76.8
Marital status	Partnered	150	71.4	70	79.6	81a	87.1
	Not partnered	52	24.8	18	20.5	12a	12.9
Branch	Army, active	172	81.9	68	78.2	74	79.6
	Army Res/Guard	32	15.2	7	5.8	7	5.4
Occupational category	Force sustainment	71	33.8	31	35.6	37	40.2
	Operations support	48	22.9	17	19.5	15	16.3
	Operations	35	16.7	18	20.7	20	21.7

Category sums may be less than total, due to exclusion of small subcategories and missing values.

^a χ ² (2, n = 96 including 3 missing) = 6.285, P = 0.043.

The mean changes in the primary and secondary outcome measures at each time point are shown on the lines in Figure 2. There were significant improvements in both PIS and composite functional score immediately following treatment. However, only the mean composite functional score improvement exceeded the MCID and showed significant sustained improvement at the 3- and 6-month follow-up. Responder analyses revealed that the percentage of study participants whose improvement met or exceeded the MCID in PIS and functional composite score was 57% and 58%, respectively, immediately following treatment; 43% and 45%, respectively, at the 3-month follow-up; and 42% for each measure at the 6-month follow-up (indicated by the bars in Fig. 1). Among participants who provided data for both the PIS and functional composite measure at 6 months, about one-third (34%) reported clinically meaningful improvement in both measures and a similar percentage (35%) reported no meaningful improvement in either measure.

FIGURE 2.

Mean pain impact and functional performance scores and percentage whose improvement exceeded the minimal clinically important difference over time.

Appendices 3, 4, and 5 show the mean scores from each PROMIS, supplemental questionnaire, and individual functional performance measure, respectively, at baseline, immediately following FR and 3 and 6 months later. The distribution of baseline PROMIS scores in the population who completed 3- and/or 6-month follow-up measures was similar to that found in the parent study.⁸ On average, mean baseline pain intensity and pain interference were in the moderate range; baseline sleep-related impairment and fatigue were in the mild range; baseline anxiety, depression, and anger were in the normal range; and physical function and satisfaction with social roles were in the average range.

Among PROMIS measures, mean pain intensity, pain interference, fatigue, and physical function showed significant improvement immediately following FR. Among the supplemental questionnaires, only the PAM and PCS reflected significant post-FR improvement. However, all outcome measures with significant improvement immediately following FR regressed toward baseline to varying degrees during the 6-month follow-up period. Although PIS improved immediately post-FR, only the pain interference component of the impact score showed sustained improvement 6 months later with a degree of improvement that exceeded the MCID.

All individual functional performance measures showed significant improvement immediately post-FR as well as 3 and 6 months later (Appendix 5). With the exception of the waist to shoulder lift, the magnitude of mean post-FR improvement exceeded the MCID for all functional performance measures immediately post-FR, but the magnitude of improvement was less than the MCID 3 and 6 months later.

Mixed regression analysis of predictors of sustained PIS improvement revealed that younger age, absence of negative neuropathic pain features, less self-reported disability, worse (higher) baseline PIS, and better (higher) baseline functional composite score predicted improvement 6 months following treatment (Table II). Analysis of predictors of functional performance improvement revealed that worse baseline functional performance, lower baseline pain catastrophizing, and more standard rehabilitation and health psychology treatment hours predicted improvement 3 and 6 months after treatment. More hours of CIH therapies did not predict sustained functional improvement (Table II). Results of mixed regression analysis of all variables are provided in Appendices 6 and 7.

TABLE II.

Mixed Regression Models of Predictors of Sustained Improvement in Pain Impact and Functional Performance

Variable	Pain Impact				Functional Performance
Demographics	n	Post-FR, b (SE)	3 months, B (SE)	6 months, b (SE)	n	Post-FR, b (SE)	3 months, B (SE)	6 months, b (SE)
Age	72	NS	NS	1.39 (0.71)*	93	NS	NS	NS
Senior vs junior military rank	71	NS	NS	NS	91	NS	5.03 (2.33)*	NS
Patient-reported measures-baseline
Neuropathic pain screen	69	NS	NS	5.12 (2.03)*	72	NS	NS	NS
Pain Impact scorea	72	−0.57 (0.13)***	−0.49 (0.13)***	−0.31 (0.14)*	72	NS	NS	NS
Pain intensitya	72	−1.25 (0.58)*	NS	NS	73	NS	NS	NS
Pain interferencea	72	−0.82 (0.18)***	−0.68 (0.18)***	NS	NS	NS	NS	NS
Physical functiona	72	0.63 (0.18)***	0.52 (0.19)***	NS	NS	NS	NS	NS
Disability self-report	47	NS	NS	0.82 (0.29)**	62	NS	NS	NS
Pain catastrophizing	67	NS	NS	NS	84	−0.22 (0.09)*	−0.19 (0.09)*	−0.26 (0.1)**
Functional performance- baseline	72	NS	NS	−0.18 (0.09)*	93	−0.35 (0.11)**	−0.57 (0.12)***	−0.32 (0.12)**
Treatment hours
Standard rehab (PT/OT) hours	72	−0.16 (0.08)*	−0.16 (0.06)*	NS	93	0.21 (0.1)*	0.21 (0.08)**	0.35 (0.08)***
CIH (chiropractic, acupuncture, yoga) hours	72	−0.28 (0.13)*	−0.28 (0.14)*	NS	93	NS	NS	NS
Health psychology hours	72	NS	NS	NS	93	NS	0.54 (0.26)*	0.83 (0.27)**

With exception of pain intensity, only variables significantly associated with 3- and/or 6 month outcomes are included in the table. Appendices 3 and 4 include all variables. All models adjusted for baseline pain impact score and composite functional score, respectively, except where noted with “a”.

^*** P < 0.001; **P < 0.01; *P < 0.05; NS = P > 0.05; PT/OT= physical/occupational therapy.

DISCUSSION

The key findings of this analysis were that 6 months following FR, significant improvement in mean functional performance but not PIS was sustained. Responder analysis revealed that 42% of participants experienced clinically meaningful improvements in PIS and/or functional performance at the 6-month follow-up. Notably, about a third of participants reported sustained improvement in both measures and another third reported improvement in neither. Given the considerable time and resource requirements of FR programs, these results have implications for selection of patients for FR participation. Among a broad range of demographic, clinical, and treatment variables studied, predictors of sustained improvement in PIS included younger age, absence of neuropathic pain features, less self-rated disability, and higher levels of baseline PIS and functional performance. Predictors of sustained improvement in functional performance included lower (better) baseline scores of PCS and lower (worse) functional performance and greater cumulative treatment hours. It is noteworthy that although the same percentage of participants reported clinically important improvement in PIS and in functional performance 6 months posttreatment, predictors of improvement were different for each measure.

Some of the earliest literature on FR identified a number of factors consistently found to predict positive response, including female sex,²³ younger age, fewer days of absence from work, having a job to return to, strong identification with the workforce, and high expectations of recovery.²⁴ More recently, Day and colleagues²⁵ identified longer pain duration to be predictive of post-FR improvement in pain intensity. Older age, neuropathic pain features, worse levels of pain intensity, and anxiety were predictive of post-FR improvement in physical function. Baseline factors previously reported to be predictive of sustained positive response following multidisciplinary pain rehabilitation include female sex,²³ history of fewer health care visits,²⁴ and higher baseline functional performance.²⁶ Posttreatment predictors of sustained response include higher levels of psychological coping.²⁷ Among studies of multidisciplinary rehabilitation for fibromyalgia, factors predictive of sustained treatment response include lower baseline use of analgesics and fewer clinical signs of central sensitization,²⁸ but baseline body mass index was not predictive.²⁹

Pain Catastrophizing

The observation that higher baseline PCS score predicted worse functional performance 6 months following FR is consistent with previous research that found PCS to be a predictor, mediator, and/or moderator of treatment outcomes. RCTs have found higher baseline pain catastrophizing to be associated with worse posttreatment disability,^30,31 pain intensity,³⁰ pain interference,³² and mental health.^31,32 Conversely, Day and colleagues found that worse baseline PCS score predicted greater posttreatment improvement in PCS score, as well as improvements in psychosocial and physical outcomes following interdisciplinary pain management.²⁵

Given the prognostic significance of baseline PCS score, it is encouraging that PCS score improved by a statistically significant 25% immediately following FR. However, the magnitude of improvement was not significant after 3-6 months. The magnitude and duration of PCS score improvement may have been limited by the relatively modest mean baseline PCS score of 20 in the current study population compared with previous studies. One study of a population of 166 occupationally disabled workers with subacute pain due to whiplash injury who participated in a 7- week pain rehabilitation program reported mean baseline and posttreatment PCS scores of 22 and 14, respectively, corresponding to an average improvement of 42%.³² Similarly, an analysis of VHA interdisciplinary pain programs reported a mean baseline PCS score range of 24-29 and an average 38% improvement immediately following treatment.⁹

Neuropathic Pain

The finding that a negative neuropathic pain screen predicted improved posttreatment PIS is consistent with the finding of Day and colleagues²⁵ that participants of an interdisciplinary pain program who did not have neuropathic pain were 4 times more likely to experience posttreatment benefit across several domains than those who did.

Cumulative Treatment Hours

The finding of an association between a higher number of cumulative treatment hours and sustained improvements in functional performance is consistent with results of a previous retrospective descriptive analysis of data collected from 882 patients who completed PASTOR at the study site between 2014 and 2018. In that study, generalized additive models showed that clinically meaningful improvement in mean PIS was observed only after the mean number of cumulative treatment hours exceeded 60.³³

Literature on the prognostic significance of treatment hours in interdisciplinary pain rehabilitation programs is conflicting, with some researchers reporting greater benefit with more hours and others finding no relationship. In a systematic review of 10 RCTs of interdisciplinary pain rehabilitation programs for chronic low back pain sponsored by the Cochrane Collaboration, Guzman and colleagues found that programs with greater than 100 contact hours yielded improved pain and function, but programs with less than 30 contact hours did not perform any better than non-interdisciplinary therapy or usual care.³⁴ More recently, the 2020 AHRQ systematic review of 6 RCTs for low back pain compared intensive (greater than 20 hours per week or greater than 80 hours total) and nonintensive interdisciplinary pain rehabilitation programs with usual care and found that nonintensive programs yielded similar improvements in pain and function as intensive programs at intermediate-term follow-up.³ Similarly, an analysis of the outcomes of 6 VHA interdisciplinary pain programs conducted by Murphy and colleagues found that program duration was not associated with outcomes.⁹

Limitations

A limitation of this analysis is that because this is a secondary analysis of both treatment groups combined, there was no comparison group to determine if outcomes were related to FR participation or to other factors, such as regression to the mean. Another limitation is the modest sample size who completed the outcome measures, although a demographic comparison of the total study population of 210 participants suggests that, with the exception of marital status, the subpopulations who completed 3- and/or 6-month measures were representative. Additionally, data for this study were collected as part of a pragmatic clinical trial, conducted during the course of usual clinical operations in a military interdisciplinary pain management center. Although participants were encouraged to adhere to study treatment and to complete study outcome measures, they were permitted to participate in the study regardless of level of adherence. This limitation was addressed by analyzing the actual number of treatment hours in which each participant engaged.

Another limitation is that the impact of other treatment approaches, such as medications and therapeutic injections which participants may have received during FR or the follow-up period, was not analyzed. Moreover, data on the type and location of pain for each participant was not analyzed, so it is unclear if the degree of response varied by pain condition. Lastly, because the inclusion criteria included minimal physical performance thresholds, the findings cannot be extrapolated to active duty service members who do not meet the functional criteria for study inclusion.

CONCLUSIONS

This analysis demonstrated the value of FR, in that over 40% of participants reported sustained clinically important improvements in PIS or functional performance and more than one-third reported improvement in both. Among the pretreatment factors identified to be predictors of sustained treatment response, pain catastrophizing³⁵ and neuropathic pain are modifiable. Further research is needed to determine if treatments that improve neuropathic pain and pain catastrophizing lead to sustained FR effectiveness.

Appendix 1. Participant Inclusion and Exclusion Criteria for the Parent Randomized Clinical Trial

 .

 

Inclusion

Exclusion

Functional impairment due to pain Able to tolerate up to 4 hours of physical activity per day without an increase in pain intensity Able to stand up and sit down on floor independently Able to walk/jog on treadmill for at least 6 minutes using modified Naughton protocol Able to tolerate at least 2 of the following: Lift at least 20 pounds from floor to waist height Lift at least 20 pounds from waist to shoulder height Carry at least 20 pounds at least 40 feet

Major surgery within 6 months before or after informed consent Unstable psychological disorder Active substance use disorder, other than tobacco High-dose opioids (≥90 mg morphine equivalent dosage per day) Not able to commit the time required for treatment

Appendix 2.

Demographics of Subsamples Who Completed PASTOR, Supplemental Questionnaires and Functional Measures 3- &/or 6-months Posttreatment

		Completed 3- and/or 6-month follow-up
		PASTOR measures (n = 91)		Supplemental questionnaires (n = 108)		Functional performance (n = 96)
Variable	Category	n	%	n	%	n	%
Sex	Male	76	84.3	91	84.3	83	86.5
	Female	15	15.7	17	15.7	13	13.5
Age	≤35 years	51	56.0	60	55.6	53	55.2
	>35 years	40	44.0	48	44.4	43	44.8
Military branch	Active Army	68	78.2	81	77.9	74	79.6
	Army Reserve and Guard	7	5.8	7	5.8	7	5.4
	Air Force	8	9.2	10	9.6	8	8.6
	Navy	6	6.9	7	6.7	6	6.5
Occupational category	Force sustainment	31	35.6	39	37.5	37	40.2
	Operations	18	20.7	23	22.1	20	21.7
	Operations support	17	19.5	18	17.3	15	16.3
	Other	21	24.1	24	23.1	20	21.7
Income	≤$50,000	34	42.0	41	42.7	35	41.2
	>$50,000	47	58.0	55	57.3	50	58.8
	Missing	10	11.0	12	11.1	11	11.5
Education	High school degree or less	21	23.6	26	24.5	22	23.2
	Some college or more	68	76.4	80	75.5	73	76.8
Marital status	Not partnered	18	20.5	21	20.0	12	12.9
	Partnered	70	79.6	84	80.0	81	87.1
Duty restriction	No or don’t know	55	61.8	65	61.3	59	62.1
	Temporary or unknown duration	20	22.5	24	22.6	23	24.2
	Permanent restriction	14	15.7	17	16.0	13	13.7
Relationship of injury to deployment	Deployment related	16	18.4	24	23.1	21	22.6
	Not deployment related	71	78.0	80	74.1	72	75.0

Appendix 3. Mean pain intensity and PROMIS measures over time.

 .

 

Appendix 4. Means scores on supplemental questionnaires over time.

 .

 

Appendix 5. Mean functional performance measures over time.

Appendix 6.

Mixed Regression Models of Predictors of Improved Pain Impact Score over Time

	N	Post-FR b(SE)	3 Months, b (SE)	6 Months, b (SE)
Demographics
Age in decade intervals	72	−0.26 (0.6)	0.6 (0.61)	1.39 (0.71)*
Female sex	72	0.71 (2.04)	0.42 (2.15)	0.19 (2.66)
Education (ordinal)	71	−1.82 (1.05)	−1.63 (1.03)	−0.25 (1.29)
Duty restriction	71	−1.71 (1.63)	−2.69 (1.66)	−2.19 (1.94)
Senior vs junior rank	71	−0.96 (1.85)	0.84 (1.89)	1.53 (2.28)
MEB status	72	2.83 (2.8)	−0.5 (2.96)	−3.22 (3.23)
Positive PC-PTSD screen	69	0.73 (2.11)	0.81 (2.21)	1.51 (2.7)
Neuropathic pain screen	69	−0.43 (1.68)	1.69 (1.72)	5.12 (2.03)*
PASTOR—Baseline
Pain impact scorea	72	−0.57 (0.13)***	−0.49 (0.13)***	−0.31 (0.14)*
Pain intensitya	72	−1.25 (0.58)*	−1.09 (0.59)	−0.5 (0.67)
Pain interferencea	72	−0.82 (0.18)***	−0.68 (0.18)***	−0.35 (0.19)
Physical functiona	72	0.63 (0.18)***	0.52 (0.19)***	0.32 (0.19)
Anger	72	−0.15 (0.09)	−0.07 (0.09)	0.02 (0.1)
Depression	72	0.1 (0.1)	0.17 (0.1)	0.1 (0.11)
Anxiety	72	−0.01 (0.1)	0.06 (0.5)	0.11 (0.11)
Fatigue	72	0 (0.11)	0.04 (0.12)	0.13 (0.12)
Sleep	72	0.08 (0.1)	0.09 (0.1)	0.13 (0.11)
Social satisfaction	72	0.19 (0.14)	0.12 (0.14)	0.02 (0.14)
Supplemental questionnaires—Baseline
Pain catastrophizing	67	−0.01 (0.07)	0.08 (0.07)	0.07 (0.08)
Pain acceptance	66	0.24 (0.15)	0.12 (0.15)	−0.1 (0.17)
Pain self-efficacy	68	0.01 (0.08)	−0.06 (0.08)	0 (0.1)
Roland-Morris disability	47	0.37 (0.26)	0.67 (0.28)*	0.82 (0.29)**
Kinesiophobia	60	0.11 (0.16)	0.08 (0.16)	0.14 (0.18)
Patient activation	71	0.06 (0.06)	0.06 (0.06)	0.07 (0.07)
Functional Tests—Baseline
Treadmill time (minutes)	72	−0.15 (0.15)	−0.1 (0.15)	−0.26 (0.17)
METS	72	−0.39 (0.34)	−0.19 (0.33)	−0.56 (0.37)
Floor-waist lift test	72	−0.28 (0.03)	−0.02 (0.03)	−0.06 (0.03)
Waist-shoulder lift test	72	−0.01 (0.05)	−0.02 (0.05)	−0.06 (0.05)
40-ft carry test	72	−0.01 (0.04)	−0.02 (0.04)	−0.07 (0.04)
Functional composite score	72	−0.07 (0.09)	−0.06 (0.09)	−0.18 (0.09)*
Treatment hours
Standard rehab	72	−0.16 (0.08)*	−0.16 (0.06)*	−0.08 (0.08)
CIH	72	−0.28 (0.13)*	−0.28 (0.14)*	−0.16 (0.16)
Health psychology	72	−0.23 (0.18)	−0.28 (0.18)	−0.02 (0.23)
Total	72	−0.1 (0.04)*	−0.1 (0.04)**	−0.05 (0.05)

All models adjusted for baseline pain impact score expect where noted with “a”. PC-PSTD = primary care posttraumatic stress disorder 5-item screen.

^*** P < 0.001.; **P < 0.01.; *P < 0.05.

Appendix 7.

Mixed Regression Models of Predictors of Improved Functional Performance over Time

	n	Post-FR, b (SE)	3 Months, b (SE)	6 Months, b (SE)
Demographics
Age in decade intervals	93	−0.18 (0.77)	0.82 (0.78)	0.72 (0.86)
Female sex	93	−7.31 (2.99)*	−4.93 (3)	−4.01 (3.09)
Education (ordinal)	92	2.39 (1.33)	2.15 (1.36)	2.11 (1.52)
Duty restriction	92	1.92 (2.18)	1.56 (2.25)	0.51 (2.38)
Senior vs junior rank	91	−0.66 (2.27)	5.03 (2.33)*	4.36 (2.58)
MEB status	93	−4.28 (3.76)	−4.59 (3.92)	−0.1 (4.86)
Positive PTSD screen	67	−0.86 (3.16)	1.65 (3.14)	1.21 (3.39)
Neuropathic pain screen	72	0.5 (2.68)	−1.17 (2.64)	−3.44 (2.82)
PASTOR—Baseline
Pain impact scorea	72	−0.16 (0.19)	−0.09 (0.19)	−0.14 (0.2)
Pain intensitya	73	−0.74 (0.81)	−0.25 (0.77)	−1.13 (0.84)
Pain interferencea	75	−0.11 (0.26)	−0.11 (0.26)	−0.09 (0.26)
Physical functiona	75	−0.15 (0.27)	−0.02 (0.27)	0.18 (0.28)
Anger	75	0.11 (0.11)	0.06 (0.11)	0.06 (0.12)
Depression	75	−0.06 (0.13)	−0.11 (0.14)	0.01 (0.14)
Anxiety	75	−0.01 (0.13)	−0.13 (0.13)	−0.01 (0.14)
Fatigue	75	−0.04 (0.15)	−0.03 (0.15)	−0.21 (0.16)
Sleep	75	−0.1 (0.14)	−0.12 (0.14)	−0.03 (0.16)
Social satisfaction	75	−0.11 (0.19)	−0.01 (0.19)	0.03 (0.2)
PASTOR Plus—Baseline
Pain catastrophizing	84	−0.22 (0.09)*	−0.19 (0.09)*	−0.26 (0.1)**
Pain acceptance	84	−0.35 (0.2)	−0.16 (0.2)	−0.09 (0.2)
Pain self-efficacy	86	0.05 (0.11)	0.07 (0.1)	0.09 (0.11)
Roland-Morris disability	62	−0.38 (0.28)	−0.29 (0.29)	−0.11 (0.29)
Kinesiophobia	80	−0.41 (0.19)*	−0.26 (0.19)	−0.34 (0.2)
Patient activation	89	−0.02 (0.08)	−0.03 (0.07)	−0.02 (0.07)
Functional Tests—Baseline
Treadmill time (minutes)	93	−0.34 (0.19)	−0.6 (0.2)**	−0.43 (0.21)*
METS	93	−0.79 (0.42)	−1.45 (0.43)**	−1.07 (0.46)*
Floor-waist lift test	93	−0.12 (0.04)**	−0.17 (0.04)***	−0.1 (0.04)*
Waist-shoulder lift test	93	−0.17 (0.06)**	−0.26 (0.07)***	−0.11 (0.06)
40-ft carry test	93	−0.14 (0.05)*	−0.25 (0.06)***	−0.14 (0.05)**
Functional composite score	93	−0.35 (0.11)**	−0.57 (0.12)***	−0.32 (0.12)**
Treatment hours
Standard rehab	93	0.21 (0.1)*	0.21 (0.08)**	0.35 (0.08)***
CIH	93	0.22 (0.16)	0.19 (0.16)	0.27 (0.17)
Health psychology	93	0.4 (0.26)	0.54 (0.26)*	0.83 (0.27)**
Total	93	0.13 (0.06)*	0.13 (0.05)**	0.21 (0.05)***

All models adjusted for baseline composite functional score except where noted with “a”. PC-PTSD = primary care posttraumatic stress disorder 5-item screen. ***P < 0.001.; **P < 0.01.; *P < 0.05.

SUPPLEMENT SPONSORSHIP

This article appears as part of the supplement “Proceedings of the 2022 Military Health System Research Symposium,” sponsored by the Assistant Secretary of Defense for Health Affairs.

FUNDING

This research was supported by grants from the Defense Medical Research and Development Program award number W81XWH-14-DMPDP-CRI-IRA-MTI and the National Institute of Nursing Research of the National Institutes of Health award number K24NR015340.

CONFLICT OF INTEREST STATEMENT

None declared.